--- name: investor_data_room description: Preparing an investor or buyer data room that survives diligence: folder structure, full document checklist by stage, gap report, naming and version rules, staged access control, and Q&A discipline. Use when building, auditing, refreshing, or closing a fundraising or sale data room, or when an investor or buyer sends a diligence request list. --- # Investor Data Room ## Purpose Build a data room a reviewer can trust and move through fast, and keep it true during a live raise or sale. The room is evidence for every claim in the deck or sale memo. A room that is complete, current, correctly gated and honestly labelled on its gaps closes diligence faster than a room with more files and no order. Output is a working room skeleton, a scored gap report, and an access plan, not legal advice. ## Workflow 1. Fix the stage and the gate first. Ask: raise or sale, what stage (seed, Series A or later, sale), who is already in, and what gate opens full diligence (term sheet, signed confidentiality agreement, or counsel set equivalent). Calibrate the checklist to that stage. Seed diligence is light by design. Scoring a seed room against a sale checklist manufactures gaps nobody asked the company to fill. 2. Create the skeleton. Run `python3 bin/data_room_audit.py init --out --stage seed|series-a|sale`. It creates the numbered folders, an `_archive` subfolder in each, an `index.csv` in `00_Index_and_Cover` with one row per checklist item for that stage, and a `gap_tracker.csv`. Folder set: 00 Index and Cover, 01 Corporate, 02 Cap Table and Equity, 03 Financials, 04 Tax, 05 Commercial and Customers, 06 Product Technology and IP, 07 Team and HR, 08 Legal and Compliance, 09 Insurance and Risk. 3. Fill against the checklist. Work through `references/data-room-checklist.md` line by line. For each line record present, missing with a fix, or not applicable with a one line reason. Signed copies only where the line says signed. Drafts labelled partial in the index. 4. Run the audit. Run `python3 bin/data_room_audit.py audit --root --stage --md-out gap-report.md --json-out audit.json`. It checks folder and checklist coverage by file name keywords, naming violations, duplicate content by hash, duplicate names, and a stale 409A warning. Exit code 1 means blocker gaps exist. 5. Do the manual checks the scan cannot do. Compare IP assignment signatories against the full founder, employee and contractor roster. Check signatures are complete on material contracts. Tie the headline numbers across deck, model and management accounts. Name the checker and the date for each in the gap report. A file being present proves nothing about any of these. 6. Set access before inviting anyone. Apply the three tier model and group rules in `references/access-control.md`: Tier 1 interest, Tier 2 diligence, Tier 3 confirmatory and sensitive. Groups per party and per role, least privilege, view only by default for external parties, one named room administrator, and a dated audit trail check cadence. 7. Write the gap report. Use `assets/gap-report-template.md`. Lead with verdict, stage, file count and coverage. Then blockers with owner and due date, other gaps, naming and duplicate fixes, access check, manual checks, and a plain list of what was not checked. Copy open items into `gap_tracker.csv` (starter rows in `assets/gap-tracker.csv`) and work that sheet until it is empty or every remaining item has a not applicable reason. 8. Run the live process from the room. Questions go in one tracked Q&A log, answered with a link to a source file, never from memory. Refresh dated financials monthly, weekly near a closing. Move superseded versions to `_archive` the same day a replacement lands. Add known problems to the disclosure memo early, in your own words. 9. Close the room properly. Export the index, Q&A log and audit trail into the closing set. Revoke external groups in tier order, Tier 3 first, confirm revocation, and set the room read only. ## Output Contract - A room build ships: the folder skeleton, a completed index, the checklist marked line by line for the correct stage, and a cover memo naming the gate for each access tier and who owns questions by workstream. - An audit ships one gap report in the template format: verdict first, blockers with owner and due date, coverage percent from the tool, and a not checked list. No verdict without the tool run and the three manual checks, or a statement that a manual check was skipped and why. - Every redaction, tier exception and not applicable line carries a reason a reviewer can read in the index. Unexplained absences are reported as gaps. ## Operating Rules 1. One authoritative version per document, in the main folder. Superseded versions live in `_archive`. File names follow the pattern in `references/access-control.md`: type, descriptor, as of date, version. The word final never appears in a file name. 2. Stage the sensitive material. Individual compensation, unredacted contracts, personal data and security detail that would help an attacker sit in Tier 3, named people only, late in the process. In a sale, check the audit trail before and after moving anything out of Tier 3 that a competitor party may already have seen, and tell counsel the same day. 3. Never backdate a document, never present in house financials as audited, never label a patent application as a granted patent. If a consent or signature was missed, fix it with a current dated ratifying document drafted by counsel and disclose it. 4. Numbers must tie. If the deck, model and accounts disagree, add a dated bridge memo that reconciles them or fix the wrong one before a reviewer finds the mismatch. 5. Disclosure beats discovery. A known issue raised by the company, with a fix plan, costs less than the same issue found by the other side. Keep the disclosure memo current. 6. This skill covers completeness, order and control. It does not value the company, interpret contracts, or give tax or legal conclusions. Route those questions to the deal lawyers and accountants, and record in the report that they were routed.